Mar 7, 2016 - evaluation and optimization of specific protocols for different types of tissues is urgently .... 3, method 2.SW had the best extraction efficiency for most metabolites in nonpolar .... thus represents a powerful tool for evaluating and optimizing ... charge on the. ACS Publications website at DOI: 10.1021/acs.anal-.
Technical Note pubs.acs.org/ac
Optimization and Evaluation Strategy of Esophageal Tissue Preparation Protocols for Metabolomics by LC−MS Huiqing Wang,† Jing Xu,† Yanhua Chen,† Ruiping Zhang,† Jiuming He,† Zhonghua Wang,† Qingce Zang,† Jinfeng Wei,‡ Xiaowei Song,† and Zeper Abliz*,† †
State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing 100050, People’s Republic of China ‡ New Drug Safety Evaluation Center, Institute of Materia Medica, Peking Union Medical College, Beijing 100050, People’s Republic of China S Supporting Information *
ABSTRACT: Sample preparation is a critical step in tissue metabolomics. Therefore, a comprehensive and systematic strategy for the screening of tissue preparation protocols is highly desirable. In this study, we developed an Optimization and Evaluation Strategy based on LC−MS to screen for a highextractive efficiency and reproducible esophageal tissue preparation protocol for different types of endogenous metabolites (amino acids, carnitines, cholines, etc.), with a special focus on low-level metabolites. In this strategy, we first selected a large number of target metabolites based on literature survey, previous work in our lab, and known metabolic pathways. For these target metabolites, we tested different solvent extraction methods (biphasic solvent extraction, two-step [TS], stepwise [SW], all-in one [AO]; single-phase solvent extraction, SP) and esophageal tissue disruption methods (homogenized wet tissue [HW], ground wet tissue [GW], and ground dry tissue [GD]). A protocol involving stepwise addition of solvents and a homogenized wet tissue protocol (SWHW) was superior to the others. Finally, we evaluated the stability of endogenous metabolites in esophageal tissues and the sensitivity, reproducibility, and recovery of the optimal protocol. The results proved that the SWHW protocol was robust and adequate for bioanalysis. This strategy will provide important guidance for the standardized and scientific investigation of tissue metabolomics.
B
is achieved by grinding in liquid nitrogen with a cooled mortar and pestle, manually degrading cold tissue with scissors,14 or homogenizing frozen tissue using an electric homogenizer;15 these methods are applied as appropriate depending on the “toughness” or “softness” of the target tissue.16−19 Extraction is usually achieved by either single-phase20 or biphasic extraction; biphasic separation has been popular in recent studies.17,21−23 The steps involving solvent addition have also been studied.10,24,25 However, few studies have systematically and comprehensively compared these various aspects of tissue preparation protocols. Evaluation of extractive efficiency remains immature, especially for low-level metabolites that play significant roles in the body and deserve close attention,26,27 and current research has focused mainly on method reproducibility. Hence, tissue preparation protocols obtained by the current approach will not provide accurate biological interpretations of data from tissue samples.
ecause tissues are the locations of most cancerous lesions,1,2 and global profiling of metabolites in tissues can reveal changes that occur during pathological conditions,3 tissue metabolomics is advantageous relative to analysis of other biological fluids (e.g., plasma/serum or urine). Consequently, tissue metabolomics is playing an increasingly important role in various fields of biomedicine.1,4−8 Because it affects not only the obtained molecular characteristics but also the biological interpretation, optimal sample preparation is key to acquiring the most accurate information.9 The choice of tissue preparation protocol is highly dependent on the type of tissue used in the research.10−13 Therefore, a systematic and comprehensive strategy for evaluation and optimization of specific protocols for different types of tissues is urgently needed. Because of its advantages in resolution, throughput, sensitivity, and specificity, liquid chromatography coupled with mass spectrometry (LC−MS) has emerged as a leading technique for global metabolite profiling data. To date, liquid extraction has played an important role in tissue preparation, which involves sample collection and quenching, tissue homogenization, and metabolite extraction. Conventionally, breakdown of the frozen tissue © 2016 American Chemical Society
Received: December 12, 2015 Accepted: March 7, 2016 Published: March 7, 2016 3459
DOI: 10.1021/acs.analchem.5b04709 Anal. Chem. 2016, 88, 3459−3464
Technical Note
Analytical Chemistry Esophageal cancer is the eighth most common malignant tumor in the world, with 456 000 new cases in 2012.28 Our group has conducted research plasma29 and urine metabolomics of esophageal carcinoma. Given the importance of tissue analysis, we have also conducted metabolomics studies of esophageal cancer tissue samples. For the reasons described above, optimization of esophageal sample preparation would be valuable for metabolomics research in this tissue. Therefore, we evaluated different methods for tissue homogenization, solvent extraction, and solvent addition in the context of LC−MS metabolic profiling of esophageal tissue, with the goal of producing a robust and efficient protocol. In this study, we established an Optimization and Evaluation Strategy for tissue preparation protocols to determine an optimal protocol for esophageal tissue metabolomics (Figure 1)
Table 1. Different Solvent Extraction and Tissue Disruption Methods for Investigation of Esophageal Tissue Preparation Protocols for Metabolomics classification of the solvent system
methods 1
solvent extraction methods
tissue disruption methods
methanol/ water;dichloromethane/ methanol
1-a Homogenized Wet Tissue (HW) 1-b Ground Wet Tissue (GW) 1-c Ground Dry Tissue (GD) 2-a Homogenized Wet Tissue (HW) 2-b Ground Wet Tissue (GW) 2-c Ground Dry Tissue (GD) 3-a Homogenized Wet Tissue (HW) 3-b Ground Wet Tissue (GW) 3-c Ground Dry Tissue (GD) 4-a Homogenized Wet Tissue (HW) 4-b Ground Wet Tissue (GW) 4-c Ground Dry Tissue (GD)
TS: two-step biphasic extraction
2
dichloromethane/ methanol/water
SW: stepwise 3
dichloromethane/ methanol/water
AO: all-in one single phase extraction
4
methanol/water 4:1
SP: single phase
consisted of aqueous 0.1% formic acid and acetonitrile, with linear gradient elution. For polar extracts, separation was performed on a Phenomenex Kinetex HILIC column (2.6 μm 2.1 mm × 150 mm) at a column temperature of 45 °C; the mobile phase consisted of aqueous 5 mM ammonium acetate and acetonitrile, with linear gradient elution. Full scan in both positive- and negative-ion modes was used for sample analysis. Detailed information about data handling and statistical analysis are provided in the Supporting Information. Method Selection. The Optimization and Evaluation Strategy mentioned above (Figure 1) was followed to evaluate and optimize different protocols for esophageal tissue preparation. To achieve a comprehensive and systematic analysis, this study focused on both extraction efficiency (for as many kinds of metabolites as possible) and reproducibility. Comparison of Extraction Efficiency and Reproducibility. Extraction efficiency of different methods for solvent extraction and tissue disruption was calculated from the peak area of the metabolites; a better tissue preparation approach will have better extraction efficiency for most kinds of metabolites. To judge the reproducibility of different protocols, we used multivariate statistical analysis, scatter plots, error bars, and relative standard deviation (RSD) of the peak area. Method Validation. Nineteen representative endogenous metabolites in esophageal tissues were selected and added to tissue samples for methodological study; standard compounds are summarized in the Supporting Information. Validation of the analytical method was performed by assessing sensitivity, precision, stability, and recovery. A detailed validation summary, including acceptance criteria, is provided in the Supporting Information.
Figure 1. Optimization and Evaluation Strategy for the screening of different tissue preparation protocols for metabolomics by LC−MS.
that could also be used in other types of tissues. First, we selected as many different representative endogenous metabolites as possible and used these target metabolites to evaluate and optimize different protocols. The most important evaluation criterias were extraction efficiency and reproducibility. After selection of the optimal protocol, we subjected the method to validation.
■
EXPERIMENTAL SECTION Sample Collection and Preparation. Rat esophageal tissues were collected from the New Drug Safety Evaluation Center, Institute of Materia Medica, Peking Union Medical College. To identify the optimal tissue preparation protocol, three kinds of tissue disruption methods and four kinds of solvent extraction methods were designed through experimental investigation. Detailed sample information and experimental design are summarized in Table 1 and in Methods, Supporting Information. LC−MS Analysis and Data Processing. Mass spectrometry experiments were performed on a Dionex UltiMate3000 HPLC system coupled to a Q-Exactive mass spectrometer with an HESI probe, controlled by the Xcalibur 2.3 software (Thermo Fisher Scientific, Waltham, MA). For nonpolar extracts, chromatographic separation was performed on a Waters ACQUITY UPLC CSH C18 column (1.7 μm, 2.1 mm × 100 mm) at a column temperature of 30 °C; the mobile phase
■
RESULTS AND DISCUSSION Metabolic profiling charts of esophageal tissue after optimization of LC−MS parameters are shown in Figure S1. Because of differences among tissues, specific tissue preparation protocols 3460
DOI: 10.1021/acs.analchem.5b04709 Anal. Chem. 2016, 88, 3459−3464
Technical Note
Analytical Chemistry
represented simply by the peak features extracted by the solvent.12,17,24,30 The protocol we finally selected in this study was more accurate because it was evaluated for a large number of target metabolites. Evaluation and Optimization of Extraction Efficiency and Reproducibility of Different Protocols. Extraction efficiency of different tissue disruption methods is calculated from the extracted peak area of the metabolites. For example, in the solvent extraction method of 2.SW:dichloromethane/ methanol/water, three tissue disruption methods (2-a.HW, 2b.GW, and 2-c.GD) were adopted separately. As shown in Figure 2A,B, for the 17 cholines in nonpolar extracts detected in positive-ion mode of ESI, 2-a.HW, 2-b.GW, and 2-c.GD had the best extraction efficiency for 16, 1, and 0 individual metabolites, respectively. Therefore, we chose 2-a.HW as the tissue disruption method for the experiments described in this section. On the basis of the principle outlined above, the three types of tissue disruption methods were compared systematically in other extracts and for other kinds of metabolites (Figure 2C−F); a comparison of the four types of solvent extraction methods is shown in Figures S2−S4. The final results are shown in Figure 2G,H, allowing comparison among three types of tissue disruption methods (Figure 2G). a.HW, b.GW, and c.GD had the best extraction efficiency for ten, one, and six individual metabolite species, respectively; among four solvent extraction methods (Figure 2H), 1.TS, 2.SW, 3.AO, and 4.SP had the best extraction efficiency for four, seven, five, and one individual metabolite species. Therefore, a.HW and 2.SW had optimal extraction efficiencies in esophageal tissues. In addition, we can compare the extraction efficiency of different methods using a heat map. As shown in Figure 3, method 2.SW had the best extraction efficiency for most metabolites in nonpolar extracts in the positive-ion mode of ESI. To compare the reproducibility of different protocols, we combined multivariate statistical analysis (Figure 4A), scatter plots (Figure 4B), error bars (Figure S5), and RSD of peak area. Statistical results are shown in Figure 4C,D. Tissue disruption methods a.HW, b.GW, and c.GD had the best
must be developed for metabolomics. Our Optimization and Evaluation Strategy gave important direction for screening of different tissue preparation protocols through the choice of target metabolites, evaluation of extraction efficiency and reproducibility of various protocols, and method validation. Choice of target metabolites. We identified a large number of target metabolites, which comprise endogenous metabolites in esophageal tissues, including amino acids, carnitines, cholines, nucleosides, and some low-level metabolites such as hormones and carbohydrate metabolites (Table 2). Table 2. Species and Amount of Endogenous Metabolites Found in Polar Extracts in Positive- and Negative-Ion Modes of ESI different sections
metabolites
polarpositive
amino acids carnitines phosphatidylcholines nucleosides low-level metabolites fatty acids amino acids
polarnegative
low-level metabolites
no. and name of metabolites 18 19 16 9 pregnenolone, androsterone sulfate DHEA sulfate, L-fucose, D-sorbitol 13 4 2-methoxyestrone, aldosterone, digoxin, D-talose, galactose 1-methylguanine, hypoxanthine 5,6-dihydrouridine, cytidine, 2-Aminophenol
The selected target metabolites were used for method selection. The targets were confirmed according to high-resolution mass data, standard compounds, or MS/MS spectrometry data; more information about them is provided in Table S1. As noted above, recent reports have focused mainly on reproducibility,10,17,20,24,30 and the objects of evaluation in these studies were usually unknown metabolites.17 Moreover, the evaluation of extractive efficiency remains immature and in some studies is
Figure 2. Comparison of extraction efficiency using 2.SW and different tissue disruption methods for nonpolar extracts in positive-ion modes of ESI. Results for the cholines are displayed as a column chart (A) and a 3D cone chart (B). (C−F) Amounts of metabolites with better extraction efficiency using different tissue disruption methods in nonpolar extracts for positive- and negative-ion modes of ESI (C, D) and in polar extracts for positive- and negative-ion modes of ESI (E, F). (G, H) Comparisons of different tissue disruption (G) and solvent extraction (H) methods. 3461
DOI: 10.1021/acs.analchem.5b04709 Anal. Chem. 2016, 88, 3459−3464
Technical Note
Analytical Chemistry
Method Validation. Sensitivity. Nineteen endogenous metabolites at different concentrations were analyzed by the C18/HILIC-(±)ESI-MS method using the Xcalibur 2.3 software to extract the ion peak signal-to-noise ratio (S/N). The results of sensitivity analysis are shown in Table S2. Seven compounds (tyrosine, proline, valine, phenylalanine, nicotinic acid, hippuric acid, and L-carnitine) could be easily detected in positive electrospray ionization mode, and the other compounds could be detected in both positive and negative ESI modes. The detection limits were of the order of 10% of the concentrations of these compounds in normal esophageal tissue. Stability. To identify reliable potential biomarkers, changes due to sample instability should be considered and removed based on time-dependent rules for individual metabolites. Several studies have investigated stability in this context.31,32 We also investigated stability after three freeze−thaw cycles. The RSD of metabolite extracted peak area was 2.1−27.1% (84.3% of metabolite peak area RSDs were less than 20%). The detailed results, which are shown in Table S6, indicated adequate freeze−thaw stability. We also investigated stability during storage at 4 °C for 1, 2, 3, and 7 days after the extracts were resuspended (Figure 5). The results showed that
Figure 3. Heat map of nonpolar extracts in positive-ion mode of ESI using four solvent extraction methods.
Figure 4. (A, B) Comparison of reproducibility evaluated by orthogonal partial least-squares discriminant analysis (A) and scatter plot of cytidine (B) in nonpolar extracts in positive-ion mode of ESI. (C, D) Reproducibility using different tissue disruption (C) and solvent extraction (D) methods.
Figure 5. Stability of samples placed at 4 °C for 1, 2, 3, or 7 days, evaluated by principal component analysis.
reproducibility for 11, 4, and 0 individual metabolite species, respectively (Figure 4C); therefore, we ultimately selected a.HW. On the other hand, method 2.SW had the best reproducibility for eight kinds of metabolites and was thus superior to the other three solvent extraction methods (Figure 4D). Many approaches for metabolite extraction from tissues focus on a small subset of known metabolites17 that tend to have similar properties. Because our aim was to perform global metabolic profiling of tissue samples, we need to extract and evaluate a wide range of metabolites from both polar and nonpolar extracts simultaneously. The results above (Figure 2H) show that method 2.SW had lower extraction efficiency than 1.TS in polar extracts but good extraction efficiency in global extracts (polar plus nonpolar extracts). Similarly, although method 1.TS had better reproducibility in nonpolar extracts (Figure 4D), 2.SW had good reproducibility in global extracts; therefore, method 2.SW is more suitable for global metabolomics analysis. Results for low-level metabolites, such as hormonal and carbohydrate metabolites, are shown in Figure 2C−F. On the basis of these data, we ultimately selected tissue disruption method a.HW, which had high-extractive efficiency both in polar and nonpolar extracts. We compared metabolite extraction efficiency with different key parameters in the manner described above. A detailed comparison is provided in Figure S6. The optimal sample preparation workflow is summarized in Figure S7.
endogenous metabolites were stable for 72 h at 4 °C. To analyze metabolic changes in greater detail, we selected different types of representative endogenous metabolites. As shown in Figure 6, fatty acids, amino acids, carnitines, nucleosides, and cholines exhibited obvious time-dependent changes at 4 °C (n = 6); however, fewer changes occurred during the first 72 h (change rate of most metabolites was
